LuxNox Dataset: Evaluating Source Camera Identification Beyond PRNU

Source camera identification is the process of matching an image to the individual camera device that acquired it. This problem can be solved using the Photo Response Non-Uniformity (PRNU) signal, widely recognised as a camera fingerprint. Recently, another camera fingerprint was studied: the Dark Signal Non-Uniformity (DSNU), which stems from differences in pixel responses in the absence of light. However, most available datasets do not contain dark frames, which are necessary for DSNU evaluation. This work describes LuxNox, a novel dataset that allows evaluation using both fingerprints. It includes 26 cameras from five different models; two of the models have ten instances each. The dataset includes dark frames, flat illuminated fields, natural images, and flat images with varying amounts of incoming light. All images are provided in both RAW and JPEG formats. Our experiments demonstrate the relevance of this dataset for both PRNU-and DSNU-based source camera identification.